Information Technology Reference
In-Depth Information
4.3 Comparing with Neuro-Immune Network Algorithms
In order to compare the algorithm with another AIS model, Neuro-immune net-
work, which combing immune system with the neural network, we have imple-
mented the proposed method with 25 datasets from TSPLIB. Table 3 makes
a comparison of the experimental results of the proposed EINET with the two
neuro-immune network algorithms, Pasti and Castro's method [19] and Masutti
and Castro's method [20], with respect to the best solutions and the average solu-
tions for 30 independent runs. In Table 3, the best results are emphasized in bold.
These result indicate that the proposed algorithm has a good capability to search
global optima. For the data sets eil51,eil76,berlin52,bier127,ch130,ch150,rd -
100,lin105, kroA100,kroA150,kroB200,kroC100,kroD100,kroE100we also can see
that EINET-TSP show better performance than other methods. However, for
some cities scale more than 100, our algorithm EINET-TSP can find best solu-
tion while the average level is not consistently better than the two algorithms.
Considering the population size, the EINET may improved by the parallelization
technique with the good suitability.
Tabl e 3.
A Comparison of the Experimental Results
Pastiand
Masuttiand
Instance OPT Castro's
Castro's
EINET
method[19] method[20]
Average Best Average Best Average Best
eil51 426 438.70 429 437.47 427 426.45 426
eil76 538 556.10 542 556.33 541 540.08 538
eil101 629 654.83 641 648.63 638 633.32 629
berlin52 7542 8073.97 7716 7932.50 7542 7612.00 7542
bier127 118282 121780.33 118760 120886.33 118970 119701.38 118282
ch130
6110
6291.77
6142
6282.40
6145
6262.75
6110
ch150
6528
6753.20
6629
6738.37
6602
6707.24
6528
lin105
14379 14702.23
14379 14400.17
14379 14551.56 14379
lin318
42029 43704.97
42975 43696.87
42834 43698.27 42537
kroA100 21282 21868.47
21369 21522.73
21333 21296.70 21,282
kroA150 26524 27346.43
26932 27355.97
26678 26934.67 26524
kroA200 29368 30257.53
29594 30190.27
29600 30202.23 29580
kroB100 22141 22853.60
22596 22661.47
22343 22246.21 22,141
kroB150 26130 26752.13
26395 26631.87
26264 26545.36 26133
kroB200 29437 30415.60
29831 30135.00
29637 30325.17 29562
kroC100 20749 21231.60
20915 20971.23
20915 20856.15 20,749
kroD100 21294 22027.87
21457 21697.37
21374 21491.00 21,294
kroE100 22068 22815.50
22427 22715.63
22395 22172.53 22,068
rd100
7910
8253.93
7947
8199.77
7982
8147.30
7910
rat575
6773
7125.07
7039
7115.67
7047
7132.23
6982
rat783
8806
9326.30
9185
9343.77
9246
9325.73
9104
rl1323
270199 300286.00 295060 305314.33 300770 303480.25 291044
fl1400
20127 21070.57
20745 21110.00
20851 21092.63 20665
d1655
62,128 71431.70
70323 72113.17
70918 72203.46 68324
Search WWH ::
Custom Search